Skip to content
CostPerPrompt

Saba API Pricing

Mistral · context window 33K · prices updated 2026-09-15

Input per 1M tokens
$0.2
Output per 1M tokens
$0.6
Cached input per 1M
$0.02
90% cheaper than fresh input

Where Saba sits on price

At $0.6 per million output tokens, Saba is a mid tier model — 3.5× cheaper than the median output price of $2.08, which puts it cheaper than 78% and pricier than 22% of the models we track. It sits 12th cheapest of the 25 Mistral models we track. Output is only 3.0× the input price, an unusually flat spread: long prompts hurt about as much as long answers, so prompt size is where the savings are.

Output tokens per $1
1,666,667
One full context fill
$0.0066
Cheaper than
78% of tracked models

Cached input is 90% cheaper. On an input-heavy workload you need roughly a 28% cache-hit rate to take 25% off the input line — reachable for chatbots and agents that resend the same system prompt and history.

What real workloads cost on Saba

These three workloads are the ones teams actually run on a mid-priced model — a $0.6/1M model is not bought for the same job as one ten times the price.

Workload Per request Per month
Support chatbot — 500 conversations/day (5K in / 1.4K out) Conversation history re-sent each turn — the classic prompt-caching win. $0.0018 $27.60
Document pipeline — 10K docs/day (3K in / 500 out) Batch-friendly: no user waiting, so a ~50% batch discount usually applies. $0.0009 $270
Small app — 1K requests/day (500 in / 150 out) The baseline most side projects actually run at. $0.0002 $5.70

Model your exact traffic in the API cost calculator — it preloads Saba with caching and batch options.

Saba price history

Saba's price has not moved since we began tracking it on Aug 2, 2026 — 44 days of stability in a market where 78 of the 334 models we track have repriced over the same period, including 7 of Mistral's own 25 models.

Change-points from our daily price snapshots (tracking since Aug 2, 2026; intraday moves between snapshots are not captured).

Saba vs Qwen3 VL 30B A3B Instruct

The closest-priced alternative from another vendor is Qwen3 VL 30B A3B Instruct (Alibaba (Qwen)) — priced within a rounding error on output, with $0.05 less per million input tokens. When two models land this close on price, the decision is quality on your own workload, not the price sheet: run 50 real requests through both and compare.

See Qwen3 VL 30B A3B Instruct pricing →

Cheaper alternatives

More Mistral models

Frequently asked questions

How much does the Saba API cost?

Saba costs $0.2 per million input tokens and $0.6 per million output tokens, with cached input at $0.02 per million (90% cheaper). That works out to roughly 1,666,667 output tokens per dollar.

What does the support chatbot workload cost on Saba?

Support chatbot — 500 conversations/day (5K in / 1.4K out) costs about $0.0018 per request and $27.60 per month on Saba. Conversation history re-sent each turn — the classic prompt-caching win.

What does it cost to fill Saba's 33K context window?

Sending 33K of input in a single request costs $0.0066 at $0.2 per million tokens — before any output. With prompt caching that same fill drops to about $0.0007 on repeat requests. This is why large context windows are cheap to advertise and expensive to actually use.

Is Saba worth the price?

Saba sits in the middle of the market (12 of 25 by price within Mistral). The honest test is a routing experiment: send the same 200 real requests to this model and to a tier below, and compare failure rate against the price gap — most teams find a majority of traffic never needed the pricier model.